AnsweredAssumed Answered

Executing Stream Segmentation with customized stream GRID

Question asked by bsulistioadi on May 17, 2013
Latest reply on May 28, 2013 by bsulistioadi

I was trying to find the topic I was about to ask in the forum archive, but couldn't find proper case and/or answer. So I make a new thread here.
I am in process of modeling a medium sized watershed in the tropics (Indonesia). The best DEM available is 90m SRTM and among some ancillary data available are: river network (from 50K topographic maps), soil type and land use (few years back).

Challenge popped out when I wanted to match the delineated subbasins as resulted from either ArcHydro or GeoHMS Terrain Preprocessing commands, both resulting not reliable junctions and stream geometry, especially in the lower elevation. For your information, the landscape of this area is highly vary from steep hills, undulating highplains until very flat alluvial riverbanks combined with - almost always - inundated areas further from river center. As I had the "good" stream network along with all river names, I intended to process the hydrological model calculation according to those rivers. So what I did was:

  1. Follow through all crucial steps in terrain preprocessing in HEC-GeoHMS (DEM Reconditioning - Fill Sinks - Flow Direction - Flow Accumulation - Stream Definition) and then I stopped here to compare the str output with my existing stream map.

  2. Learning that the str Grid only contains "1" value for the main river where all the water should accumulate, I then edit (include generalized all meanders/curves) my existing stream map to match the starting point and all junctions defined by Hydro/Geo-HMS, then create field called "VALUE" and assign all of them with "1", then convert it to raster and renamed it to str1

  3. Assuming the file structure already been matched between the two (str : arc defined stream and str1 : edited existing stream) I then continue the terrain pre-processing with Stream Segmentation for "str1" grid that I just created. FYI the existing stream polyline contains ONLY 58 segments/feature before I converted to grid.

  4. When investigating the result, I surprised since the str1lnk resulted from stream segmentation based on the "real" stream gives me very large value range (from 1 to 9970) compared to the strlnk grid resulted from artificial stream (from 1 to 67)

  5. Whenever I continue with the next steps (Catchment Grid Delineation - Catchment Polygon Processing - Adjoint Catchment Processing - etc) GeoHMS always gave me TONS of small catchments, which makes it difficult to manually merge them in GeoHMS

The questions are:

  • Have anybody ever tried this approach before?

  • If this approach is not a good one, how to modify the sub-basins resulted from automatic process in Hydro/GeoHMS according to the real ones?

For anyone interested to help and want to try to run the process with the dataset, I put them in dropbox linked below:

Thank you in advance

Budi Sulistioadi